But do we need universal grammar ? Comment on Lidz

“Noun phrase number is a privileged source of information as to the semantic structure of predicates.” So concludes a recent article by Lidz, Gleitman, and Gleitman (2003: 169), which purports to demonstrate the viability of a “universalist” mapping between syntax and semantics as opposed to an “emergentist” view of the mapping, attributed to Goldberg (1999) and Tomasello (2000). Lidz et al. (2003) performed an experiment involving the Dravidian language, Kannada, based on the methodology of Naigles, Gleitman, and Gleitman (1993). They presented to children (mean age 3;6) familiar verbs in familiar and unfamiliar (ungrammatical) contexts. Unfamiliar contexts included intransitive verbs presented in transitive frames and/or with causative morphology, and transitive verbs presented in intransitive frames with or without causative morphology. Children were then encouraged to act out a scene corresponding to the sentence they had heard using a set of toy animals. Of particular interest is the degree to which the children, when faced with an intransitive verb in an ungrammatical context, relied on causative morphology as compared with the transitive syntax in their interpretations. The authors observe that in Kannada, the causative morpheme is reliably associated with a causative interpretation. They further note that the transitive construction involving the overt expression of two arguments is associated with a wider range of interpretations than causation (as is also the case in English—cf. transitive clauses with the verbs know, see, want, owe). They suggest that any “emergentist” theory that claims that argument structure is learned from the input, would predict that subjects should rely more on the causative morpheme as